PRISM: Probabilistic and Robust Inverse Solver with Measurement-Conditioned Diffusion Prior for Blind Inverse Problems

2025-09-19

AI Summary

This work addresses the blind inverse problem in computational imaging where the forward operator is unknown. We propose a probabilistic, robust solution framework grounded in measurement-conditioned diffusion priors. Our key contribution is the first integration of a measurement-conditioned diffusion model into a Bayesian posterior sampling framework, enabling joint uncertainty modeling and co-optimization of the latent image and degradation kernel. Crucially, the denoising process is dynamically guided by observed measurements, eliminating reliance on an accurate forward model. Evaluated on blind deblurring, our method substantially outperforms existing state-of-the-art approaches, achieving simultaneous improvements in both image reconstruction fidelity and blur kernel estimation accuracy. Extensive experiments demonstrate its effectiveness, robustness to model mismatch, and strong generalization across diverse blur types and noise levels.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Reasoning under Uncertainty: Probabilistic Inference
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Responsible Web: Measurement, analysis, and circumvention of Web censorship

Abstract

Diffusion models are now commonly used to solve inverse problems in computational imaging. However, most diffusion-based inverse solvers require complete knowledge of the forward operator to be used. In this work, we introduce a novel probabilistic and robust inverse solver with measurement-conditioned diffusion prior (PRISM) to effectively address blind inverse problems. PRISM offers a technical advancement over current methods by incorporating a powerful measurement-conditioned diffusion model into a theoretically principled posterior sampling scheme. Experiments on blind image deblurring validate the effectiveness of the proposed method, demonstrating the superior performance of PRISM over state-of-the-art baselines in both image and blur kernel recovery.

Problem

• Solving blind inverse problems in imaging
• Addressing unknown forward operator issues
• Improving image and blur kernel recovery

Innovation

• Measurement-conditioned diffusion prior for blind problems
• Theoretically principled posterior sampling scheme
• Robust solver without complete forward operator knowledge